The Byzantine Generals Problem is a thought experiment from a 1982 paper by Leslie Lamport, Robert Shostak and Marshall Pease. The setup goes like this. A group of generals from the Byzantine army are camped outside an enemy city. They have to agree on a coordinated attack or retreat. They can only communicate by messenger. Some of the generals may be traitors who will try to sabotage the plan. How can the loyal generals reach a consistent decision even though they do not know who is lying?
The paper proved that this problem is solvable only under certain conditions, specifically that fewer than one-third of the generals are traitors. Any protocol that guarantees consistency under those conditions is said to be Byzantine Fault Tolerant (BFT). This is a foundational result in distributed computing and it was studied for decades before anyone found a practical use for it at internet scale.
Why It Matters for Crypto
Bitcoin is, at its heart, a solution to a version of the Byzantine Generals Problem. The “generals” are the nodes on the network. The “attack or retreat” decision is whether to accept a given block as the next valid state of the ledger. Some nodes may be malicious — trying to double-spend, trying to censor transactions, trying to fork the chain. The network needs to reach agreement anyway.
Satoshi Nakamoto’s 2008 paper was the first time anyone had proposed a practical BFT solution that worked without knowing who the participants were in advance. Classical BFT algorithms (PBFT and its successors) require a fixed, known set of participants. Bitcoin’s innovation was to use proof-of-work as a way to choose a leader for each block — whoever solves the hash puzzle first gets to propose the next block — which is a probabilistic solution to the same problem that scales to open, permissionless networks.
Later chains use different approaches. Ethereum, Solana, Cosmos, Tendermint and the various proof-of-stake chains all use some variant of classical BFT (usually Tendermint-based or HotStuff-based) now that they have explicit validator sets. But the underlying problem — reach consistent agreement even when some participants are adversarial — is the same problem Bitcoin solved first, with a more elegant (and more expensive) approach.
Why This Language Is Used
You will mostly hear “Byzantine fault tolerant” in the context of proof-of-stake chains explaining why their finality is safe under the assumption that fewer than one-third of validators are malicious. The threshold comes directly from the Lamport paper, and any consensus protocol that tries to do better than that is mathematically impossible in the general case. When a chain describes itself as “Byzantine fault tolerant with one-third adversarial stake”, it is telling you exactly which academic result it is relying on.